Vision2DCS is an industrial AI prototype that turns P&ID interpretation into a structured engineering workflow. It analyzes instrumentation diagrams, drafts tag mappings, and produces visual topology concepts for modern DCS and HMI projects.
Demo: YouTube walkthrough
Instrumentation engineers still spend significant time translating P&IDs into tag lists, control structures, and HMI drafts by hand. Vision2DCS demonstrates how multimodal AI can accelerate the early design phase of automation projects while keeping the output visible and reviewable.
- Parses uploaded process diagrams with Gemini-powered multimodal analysis.
- Extracts instrumentation concepts into structured engineering data.
- Drafts HMI and topology views with React Flow components.
- Maps output toward DCS-centric thinking for platforms such as Siemens PCS 7 and ABB 800xA.
- Packages the result as an explorable React/Vite prototype for portfolio review.
- Frontend: React 19, TypeScript, Vite
- AI layer:
@google/genaiwith Gemini-based image and text reasoning - Visualization: React Flow for topology and node mapping
- Reference docs:
TECHNICAL_GUIDE.mdfor deeper implementation notes
- Node.js 20+
- npm
- A Google AI Studio API key
npm install
cp .env.example .env.localnpm run devnpm run buildservices/geminiService.tscontains the AI orchestration layer.components/HmiReactFlowView.tsxdrives the topology visualization.TECHNICAL_GUIDE.mddocuments the engineering intent in more depth.
This repository is a concept-stage engineering tool. It is intended to showcase Industrial AI workflow design for automation engineering rather than serve as a production-ready export pipeline.
MIT